Effectiveness of a clinician intervention to improve physical activity discussions in underserved adults

Abstract

Background.

Physical activity (PA) counselling is challenging in primary care. It is unknown whether clinician training on the 5As (Ask, Advise, Agree, Assist, Arrange) improves PA counselling skills.

Objective.

To evaluate the effect of a clinician training intervention on PA counselling for underserved adults using the 5As framework.

Methods.

Pragmatic pilot clinical trial was used in the study. Clinicians (n = 13) were randomly assigned to two groups. Each group received the intervention consisting of four 1-hour training sessions to teach the 5As for PA counselling. Patient–clinician visits (n = 325) were audio recorded at baseline, immediately post-intervention, and at 6 months. Outcomes were the frequency and quality of PA discussions using the 5As, assessed by blinded coders.

Results.

Patients’ mean age was 44 years; 75% were African American. PA was discussed in 37% (n = 119) of visits overall and did not change from baseline to follow-up. When PA discussions occurred, the frequency of 5As increased from baseline to follow-up for Advise (51–54%), Agree (11–26%), and Assist (11–17%); however, none of the 5As had a statistically significant increase. For Agree, exploration of patient willingness to engage in PA increased from 23% at baseline to 50% at follow-up.

Conclusion.

A clinician-directed intervention to improve PA counselling increased the frequency of Advise, Agree and Assist, and the quality of Ask and Agree statements, though the absolute numbers were small and only Agree reached statistical significance. Future research is needed to understand the factors that affect the optimal uptake and approach to 5As counselling.

Introduction

Fifty percent of deaths in the USA are preventable (1), with the majority related to behaviours such as physical activity (PA), smoking, diet and the social environment (2). Primary care is challenged to translate promising PA interventions into practice, especially for promoting PA among underserved or disadvantaged groups (defined herein as members of ethnic or racial minority groups or persons with low income). Disadvantaged groups are less likely to have sufficient PA according to recommended guidelines (3) and are more likely to suffer a greater burden of disease related to inadequate PA (4). Lack of sufficient resources, greater competing demands, and multiple risk factors and chronic conditions in their patients compound these challenges (5).

A recent meta-analysis of 15 randomized controlled trials from seven countries showed that primary care-based PA counselling yielded small to medium effect sizes on patient reported PA at 12 months (6). The number needed to treat (NNT) for an intervention for one additional sedentary adult to meet the recommended guidelines for PA was 12, which compares favourably to NNT for smoking cessation advice of 50–120 (6).

The 5As framework, in which clinicians Ask about (or Assess), Advise about, Agree upon, Assist and Arrange follow-up regarding patients’ behaviour change efforts (7), is a set of guidelines for brief counselling. The 5As have been tested via randomized interventions to change PA (8), weight loss (9), dietary change (10), smoking cessation and lipid management. The 5As have produced small-to-moderate improvements in self-reported and objectively measured behaviour change. However, there are no data that demonstrate that physician training on use of the 5As actually changes their communication with patients about behaviour change.

The objective of this pilot study was to provide evidence about the feasibility of a clinician training intervention designed to increase the frequency and quality of 5As discussions for PA between primary care physicians and their low income, ethnically diverse adult patients. We also sought preliminary evidence of an effect of the intervention on patient–physician communication in order to plan a larger pragmatic clinical trial.

Methods

The study design and protocol has been previously published (11); we briefly report key features here. The study was approved by the University of Rochester’s Research Subjects Review Board and the federally qualified health centre’s Health Services Committee.

Study setting and design

All clinicians and patients were recruited from a federally qualified health centre in Rochester, NY. The health centre serves ~14000 patients who are predominantly low income and ethnically diverse. The design was a pragmatic pilot clinical trial.

Clinician recruitment and enrolment

Clinicians were recruited in person. They provided informed consent to complete a 10-minute written baseline survey, attend training sessions, have 25 office visits recorded over 3 time points for Group 1 (baseline, immediately post-intervention and 6-month follow-up) and 4 time points for Group 2 (2 baseline assessments, one coinciding with Group 1’s baseline assessment and one preceding Group 2’s intervention, plus the other assessments) and complete a survey and individual interview at the end. After informed consent, clinicians were randomly assigned to receive immediate training (Group 1) or training 8 months later (Group 2).

Patient recruitment and enrolment

Patient eligibility was determined by a research assistant using a checklist of inclusion/exclusion criteria, as reported previously (11). In summary, patients were eligible if they were adults (age ≥ 18 years), presenting for a routine health maintenance or follow-up medical appointment for which PA counselling would be appropriate, and able to provide written informed consent. Patients were excluded if they had already participated and thus did not take part in baseline and follow-up visits.

Intervention

The goal of the intervention was to increase the use of the 5As for PA counselling using interactive clinician communication training techniques. Clinicians attended three 1-hour training sessions and one individually based session. The sessions covered the following topics: (i) evidence-based recommendations for PA; (ii) definition of the 5As; (iii) eliciting motivation; (iv) problem-solving, strategies to provide support and goal setting; and (v) options to refer patients to community resources. In the fourth session, clinicians met individually with a standardized patient to assess their skills using a pre-developed 5As competencies checklist. Standardized patients are actors trained to portray patients based on pre-developed case vignettes. For session 4, standardized patients conducted role plays with each individual clinician and gave them feedback on their use of the 5As for PA counselling. To encourage clinician adoption of Arrange, a partnership with a community exercise programme was established, which allowed clinicians to refer patients to the program free of charge.

Assessment/description of audio recording procedure

When potentially eligible patients entered exam rooms for their visit, a nursing assistant mentioned the project. If interested, a research assistant obtained written informed consent. As with the clinician informed consent procedure, patients were explicitly informed that if they agreed to participate, a recorder would be placed unobtrusively in the exam room. Therefore, neither clinicians nor patients were blinded to the presence of the recorder.

Coding

All recorded office visits were transcribed verbatim and separated into two categories, based on the presence or absence of PA discussion. All transcripts containing any discussion of PA were subsequently coded for the presence and quality of the 5As. Five individuals (three undergraduate students, a PhD-level fellow and the study coordinator who trained and supervised all coders) performed the coding. Any discrepancies in the coding were resolved through discussion with the principal investigator.

The coding template was based on a previously developed coding scheme and adapted to include items from the 5As competency checklist used for the intervention. The threshold concordance goal was set to 90% or greater. Concordance was calculated for each of the four coders, relative to study coordinator. The average concordance was 91%. All coders were blinded to clinician and time point.

Data analysis

We used pre/post-analysis for Groups 1 and 2 to evaluate the impact of the training on clinician use of 5As. We calculated the frequency of the 5As by summing each A overall and across each time point. The quality ratings were categorical variables, so we calculated the percent change for each of the quality rating variables. A mixed effect model was used to compare the mean 5As score between and within each group for the three time points: baseline, immediately post-intervention and 6-month follow-up. The models included a random effect (clinician) and fixed effects (intervention group, time and baseline 5As score). An interaction term between treatment and time was used to determine if the effect of treatment changed over time.

Data were analysed using STATA version 12.1 (Copyright © 1985–2011 by StataCorp LP, College Station, TX).

Results

Participant characteristics

There were 859 adult patients assessed for eligibility, and of those, 325 (37.8%) enrolled and had recorded office visits that were transcribed and coded. The most common reasons for ineligibility were no-shows or cancellation, language barrier, acute medical condition, patient refusal and clinician or research staff time constraints. The majority of participants were female (71%, n = 225) and African American (69%). Participants’ average age was 43 years. Educational attainment was generally low and 64% of participants reported an annual income of less than $20000.

The 13 clinicians were primarily female (75%) and white (67%), with an average age of 50 years. The mean number of years in practice was 15. Nine were physicians, two were nurse practitioners and two were physician assistants.

Frequency of 5As for physical activity discussions

Of the total sample across all time points (n = 325), PA was discussed in 37% (n = 119) of recorded visits and 13% of all visits (n = 859) initially assessed for eligibility. The overall percentage of discussions did not change from baseline to follow-up. There were 45 discussions at baseline, 39 at post-intervention, and 35 at follow-up. Table 1 shows patient social demographic and other information for the 119 visits with PA discussions. Of these visits, 114 (96%) contained an Ask statement, 55 (46%) had Advise, 20 (17%) Agree, 19 (16%) Assist and 7 (6%) Arrange.

Table 1.

Demographics

	Clinicians (n = 10)	Patients (n = 119)
	% (n)	% (n)
Age (mean, SD)	47.5 (13.6)	44.5 (13.5)
Race
Black or African American	30.0% (3)	65.0% (76)
White	50.0% (5)	21.4% (25)
Other	20.0% (2)	4.3% (5)
Hispanic	0.0% (0)	9.4% (11)
Female (%)	80.0% (8)	69.2% (81)
BMI (mean, SD)	–	34.1 (9.55)
Number of years in practice (mean, SD)	15.7 (11.3)	n/a
>2 comorbidities	n/a	57.1% (68)

BMI, body mass index; n/a, not applicable; SD, standard deviation.

The frequency of 5As discussions increased from baseline through follow-up for Advise (51–54%), Agree, (11–26%) and Assist (11–17%) and Arrange (4–6%). Such increases did not occur for Ask (96–94%) (see Table 2).

Table 2.

Frequency of 5As: baseline, post-intervention and 6-month follow-up

		Baseline	Post	6-Month post	P-value
Total (n)	119	45	39	35
Ask	95.8%	95.6%	97.4%	94.3%	0.86
Advise	46.2%	51.1%	33.3%	54.3%	0.14
Agree	16.8%	11.1%	15.4%	25.7%	0.21
Assist	16.0%	11.1%	20.5%	17.1%	0.49
Arrange	5.9%	4.4%	7.7%	5.7%	0.89

Table 3 shows individual clinical level PA discussions. As Table 3 shows there was considerable variability between clinicians and time points in their individual frequency of PA discussions, ranging from a total of 0 to 16 PA discussions over all time points of the study.

Table 3.

Physical activity discussions by clinician

Clinician	Baseline	Immediate post	Follow-up	Total physical activity discussions per clinician
1	3	2	5	10
2	1	3	5	9
3	7	4	0	11
4	4	6	4	14
5	3	5	1	9
6	4	5	7	16
7	5	3	6	14
8	7	0	0	7
9	4	2	3	9
10	1	0	0	1
11	0	6	0	6
12	4	3	4	11
13	0	0	0	0

Quality of physical activity discussions

For Agree, exploration of patient willingness to engage in PA (e.g. probing for details about motivation to change) increased from 23% at baseline to 50% at follow-up. For discussions containing Assist, having a goal being set or discussed increased from 8% at baseline to 36% at follow-up; discussion about specific strategies increased from 0% at baseline to 9% at follow-up. Of all the quality ratings, the only statistically significant was the Assist rating for the clinician’s attempt to reach consensus on a PA goal, 0% at baseline to 6% at follow-up (P = 0.035). There were no changes for the Advise or Arrange quality ratings.

Discussion

Findings from this pilot study show that a clinician training intervention did not change the overall percentage of PA discussions; there was considerable variability at the level of the individual clinicians. When PA discussions did occur, there was a trend towards an increase in frequency and quality for some aspects of PA counselling increased for some of the As: Asking, Agreeing and Assisting. To our knowledge, this is the first intervention for PA counselling, which used recorded visits and blinded coding to assess the effect of the training on actual clinician PA counselling skills using the 5As.

On the basis of the results, the intervention was efficacious in improving some clinician communication skills, but not others. The skills that improved were clinicians’ attempts to explore patient willingness or motivation for change, detailed inquiry about current PA levels and attempts to help patients with goal setting. These findings are encouraging since these skills are associated with improved patient recall of discussions about behaviour change (12), patient motivation to change (13–15) and actual change attempts (12,13). These skills are also congruent with Motivational Interviewing techniques and have been shown to improve PA outcomes in several systematic reviews and meta-analyses (16,17).

Counselling skills that did not change (but that were emphasized in the intervention) were tailoring advice (Advise) to the patient, setting specific goals, mentioning community resources and Arranging follow-up. Others have documented that Arrange is the least frequently occurring A (18,19). Rather than communication training per se, knowledge about and access to low cost or free community resources for PA may best promote Arrange, which others have shown. Future research should evaluate the delivery of the 5As from different clinical team members and community partners and also assess the 5As using multiple data sources.

Despite the observed changes in the frequency and quality of some of the As, the overall percentage of visits containing any discussion of PA did not change during the study. The overall percentage of visits containing any discussion of PA (about one-third of all visits), is similar to other findings (19). Though we did not expect PA to be discussed in every visit, we did hypothesize that the frequency of discussions would increase. Our findings may be due to a combination of reasons related to the barriers to clinician counselling commonly reported (20), which our clinicians acknowledged as well: the difficulty of changing, competing demands and a challenging patient population with numerous socio-economic and health-related barriers to exercise. Patients in this study reported significant health-related barriers to PA, yet also reported many sources of motivation, support and interest being referred to community programs for PA (21). Given the level of clinician variability in PA discussions we observed, it is possible that some clinicians were more comfortable immediately adopting the 5As into their counselling than others. Our study was unable to offer booster sessions or other forms of support for clinicians to augment their 5As skills for those that may have needed more time to feel comfortable using the 5As. Therefore, future research should endeavour to tailor PA counselling and referral to both build on motivation and address health and socio-economic circumstances; this will require effort, teamwork and partnerships beyond the 15-minute office visit.

Strengths of the study include use of recorded visits, a mix of clinician types, focus on an underserved population, a robust coding scheme and use of multiple data sources. However, generalizability is limited because the study was conducted in only one setting. Also, this pilot project did not permit us to collect PA outcomes from participants.

Conclusions

Pilot data suggest that a clinician-directed intervention to improve PA counselling can increase the frequency of Advise, Agree and Assist, and the quality of Ask and Agree statements. Future research is needed to understand the factors that affect the optimal uptake and approach to 5As counselling.

Declaration

Funding: National Cancer Institute of the National Institutes of Health (Award Number K07CA126985).

Ethical approval: University of Rochester’s Research Subjects Review Board and the federally qualified health centre’s Health Services Committee.

Conflict of interest: none.